It seems to me that he is singing the praises of lanagues which are (according to your definitions) both strongly typed and statically typed.
His Spectacular Example shows ML detecting an infinite loop at compile time (ie statically) but ML's fancy static type system wouldn't be much better than Java's if it wasn't also strong (that would essentially be Java + type inference, allowing you to cut out lots of declarations but still leaving you open to run time type errors). Strong without static would be Perl and that talk is definitely not a celebration of Perl's type system.
The definitions are controversial and somewhere (possibily in that or another MJD talk) there's a list of about 8 conflicting definition for various kinds of typing but in that talk, on this page he says "static" and "strong" typing are the same.
If you really want to make a distinction you could take "strong" to mean that conversions (casts) are impossible and "static" to mean that types are determined at compile time. Then, if you want to say that Java really is statically typed you should probably make it clear that you mean that "types are determined at compile time but might have been determined incorrectly". Come to think of it, that's a great definition, it makes it clear just how badly Java, C, and C++'s type systems suck.
Maybe we should call Java a "hopefully typed" language :-).
<code> <a> <b> <big> <blockquote> <br /> <dd> <dl> <dt> <em> <font> <h1> <h2> <h3> <h4> <h5> <h6> <hr /> <i> <li> <nbsp> <ol> <p> <small> <strike> <strong> <sub> <sup> <table> <td> <th> <tr> <tt> <u> <ul>